Sentiment and Other Content Analyses

Loading Packages

library(tidyverse)
library(tidytext)
library(gutenbergr)

Installing a package

# if(!require("topicmodels")){
#   install.packages("topicmodels", type = "binary")
#   library(topicmodels)
# }

if(!require(slider)){
  install.packages("slider")
  library(slider)
}

if(!require(textdata)){
  install.packages("textdata")
  library(textdata)
}

frank <- gutenberg_download(84)

frank |> 
  mutate(                
    text = str_squish(text),
    section = case_when(
      str_detect(text, "^Letter \\d+$") ~ text,
      str_detect(text, "^Chapter \\d+$") ~ text
    )
  ) |>             
  fill(section) |>
  filter(text != section,
         !is.na(section),
         str_length(text) > 0) |>
  mutate(line_number = row_number()) ->
  frankenstein

1

Remove whitespace (spaces, tabs, newlines) from the beginning and end of lines

2

If a line has “Letter 1” or “Chapter 1” etc, copy it to the column section

3

Fill values in section downwards

4

Drop lines where the text and the section are the same

5

Drop lines which don’t have a section (header of the book)

6

Drop lines which are empty.

7

Add line numbers.

Length of each chapter in lines

frankenstein |> 
  group_by(section) |>
  summarise(
    start = min(line_number),
    end   = max(line_number)
  ) |>
  mutate(length = end - start) |>
  separate_wider_delim(
    section,
    delim = " ",
    names = c("prefix", "number"),
  ) |>
  ggplot(aes(as.numeric(number), length))+
    geom_col(fill = "grey90")+
    facet_grid(
      .~prefix,
      space = "free_x",
      scales = "free_x"
    )

1

x

2

x

3

x

4

x

5

x

Sentiment Analysis

Preparing for “Sentiment Analysis”

frankenstein |> 
  unnest_tokens(
    word,
    text,
    token = "words"
  ) |> 
  mutate(
    word = str_remove_all(word, "_")
  ) ->
  frank_words

“Sentiment Lexicons”

get_sentiments(lexicon = "bing") |> 
  count(sentiment)

# A tibble: 2 × 2
  sentiment     n
  <chr>     <int>
1 negative   4781
2 positive   2005

get_sentiments(lexicon = "afinn") |> 
  count(value)

# A tibble: 11 × 2
   value     n
   <dbl> <int>
 1    -5    16
 2    -4    43
 3    -3   264
 4    -2   966
 5    -1   309
 6     0     1
 7     1   208
 8     2   448
 9     3   172
10     4    45
11     5     5

get_sentiments(lexicon = "loughran") |> 
  count(sentiment)

# A tibble: 6 × 2
  sentiment        n
  <chr>        <int>
1 constraining   184
2 litigious      904
3 negative      2355
4 positive       354
5 superfluous     56
6 uncertainty    297

get_sentiments(lexicon = "nrc") |> 
  count(sentiment)

# A tibble: 10 × 2
   sentiment        n
   <chr>        <int>
 1 anger         1245
 2 anticipation   837
 3 disgust       1056
 4 fear          1474
 5 joy            687
 6 negative      3316
 7 positive      2308
 8 sadness       1187
 9 surprise       532
10 trust         1230

Adjoining data

frank_words |> 
  left_join(
    get_sentiments(lexicon = "afinn")
  ) -> 
  frank_word_sentiments

Summary by section

frank_word_sentiments |> 
  group_by(section) |>
  summarise(
    value = mean(value, na.rm = T),
    starts = min(line_number),
    ends = max(line_number),
    midp = median(line_number)
  ) ->
  sentiment_by_section

sentiment_by_section |> 
  ggplot(aes(starts, value))+
    geom_line(aes(x = midp))+
    geom_segment(
      aes(
        xend = ends,
        yend = value,
        color = value
      ),
      linewidth = 5,
      lineend = "round"
    )+
    khroma::scale_color_bam()

1

x

2

x

3

x

4

x

Sliding mean

Sliding window

slide_dbl(
  1:10,
  .f = ~mean(.x),
  .before = 2,
  .end = 2
)

 [1] 1.0 1.5 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0

What about chapter 24

frank_word_sentiments |> 
  group_by(section) |> 
  filter(
    section == "Chapter 24"
  ) |> 
  mutate(
    word_order = row_number(),
    sliding_value = slide_dbl(
      value,
      .f = ~mean(.x, na.rm = T),
      .before = 500,
      .after  = 500
    )
  ) |> 
  ggplot(aes(word_order, sliding_value))+
    geom_line(
      aes(color = sliding_value)
    ) +
    khroma::scale_color_bam()